You get an order from a building company to make one ton of nails. What will you do? I guess you’ll produce big nails, for that’s easier and cheaper than producing small nails. From another building company you get an order for one million nails. For them, you’ll produce little nails, I think, for you need less iron to make them. If you behave that way, you follow Goodhart’s Law. Such behaviour is not a figment of my imagination, but it often happened, for example, in the former Soviet Union in order to fulfil to targets imposed by the planning authorities.
Goodhart’s Law was developed in 1975 by Charles Goodhart. He formulated it this way:
Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes.
Later Marilyn Strathern gave it its present wording, which is generally used since then:
When a measure becomes a target, it ceases to be a good measure.
Strathern gave the law also its present name. Applied to the introductory case of this blog: The nail factory gets an order but doesn’t ask itself what the customer needs, such as nails of different shapes and sizes. It only wants to execute the order as it is, whether that’s reasonable or not. The customers can avoid such a reaction by specifying the types of nails they need as precise as possible.
Initially, Goodhart’s Law was used to describe economic behaviour. Later it got also a political interpretation and actually it can be applied, if relevant, to any kind of actions.
At the basis of this law is a general sociological phenomenon that has been described by Jürgen Habermas in his theory of two levels of meaning: level 1 and level 0. Level 1 is the level all sciences are faced with when they theoretically interpret their objects of research. Level 0 is typical of those sciences that deal with objects that have been given meaning by the investigated people themselves. This made me distinguish two kinds of meaning: meaning 1 and meaning 0. (see my 2001 for a detailed explanation) Meaning 1 is the kind of meaning used on level 1. It is the meaning a scientist gives to an object, either physical or social in character; it is the scientist’s theoretical interpretation of reality. Meaning 0 is the concept of meaning for the underlying level 0. It is the meaning people who make up social reality give to this social reality or to parts of it themselves; it is their interpretation of their own lived reality. We can apply this two-layer model also to Goodhart’s Law. Then we get this: Level 1 is the level at which the target is set in objective (measurable) quantities, for example by a customer or a policy maker. The people who have to realize the target are at level 0. They give it their own interpretation; one that fits them best. However, this subjective interpretation does not need to be what those who formulated the target thought it should be. If the people who have to realize the target literally take it as it is, it’s quite well possible that the target ceases to be a good measure. If so, Goodhart’s Law applies.
Goodhart’s Law can occur everywhere where quantitative targets are set, and so we see it also appear in the present Covid-19 pandemic. Quantitative targets are set to indicate when the pandemic has become manageable and everything is done to reach them. However, at the same time the negative effects of the measures to achieve the targets are not seen, ignored or pushed away as not important in view of the higher goal to reduce the number of infections. Or the targets are simply avoided by the people by changing their behaviour. As for the latter, for instance, people avoid the curfews imposed by meeting others or by shopping at hours that it is allowed to do so, which leads to more people being together at other moments. As for the former, everywhere all is done to reduce the number of Covid-19 patients, and as such it is a good target, but the price is high. Patients with other serious diseases and illnesses often cannot get the treatments they need, for instance because there is no room for them in the hospitals; in many cases this has led to early deaths. Or, another example, people have died from loneliness in nursery homes and at home because they were and often still are not allowed to meet other people, especially their relevant others. Again other people come to suffer of depressions, often so seriously that they commit suicide. Children suffer because they cannot go to school. Or think of the big negative consequences the restrictions have for the economic lives of many people, making that their quality of life has gone down and probably will remain so for many years.
Now I am the last to say that we should take less care of the Covid-19 patients and that we shouldn’t take measures to stop the pandemic. That’s not the point. The point is that at the moment only the pandemic counts, and politicians pay attention to the negative effects of the restrictions only with words but not with deeds. It’s striking, for instance that the Dutch Outbreak Management Team, which has the task to inform the Dutch government about the pandemic and to propose measures to contain it, has only doctors, virologists and the like among its members but no psychologists and sociologists, who could assess the social and psychological effects of the measures and propose alternatives. Then one must not be surprised that Goodhart’s Law applies and that the anti-Covid measures cease to be good measures.
Sources
- Henk bij
de Weg, “The
commonsense conception and its relation to philosophy”, Philosophical
Explorations, 2001/1, pp. 17-30.
- “Goodhart’s Law, in Wikipedia, https://en.wikipedia.org/wiki/Goodhart%27s_law
No comments:
Post a Comment